n.V6-4.02 | BUILDING IT
01| Building IkoCiti means producing a global digital infrastructure — layers of technology, a set of marketplaces, training systems, data pipelines, compliance scanning, the PI Exchange. Code, architecture, integration, security, governance. That is the object.
02| A meaningful slice of that work is genuinely absorbed by AI today. Code generation is real. Scaffolding is real. Documentation production — the part every technical team hates and skips — runs faster with AI than without. Test generation, security scanning, and integration boilerplate come out at close to human quality for a fraction of the cost. Architecture decisions stay human, but architecture exploration does not: generating options, comparing trade-offs, and pulling patterns from comparable platforms is exactly what the machine is good at. A small team with current tools produces in a quarter what a large team produced in a year not long ago.
03| That is the strong half. The weak half is not a temporary limitation waiting for the next model.
04| AI does not produce the integration logic that makes a platform coherent rather than a pile of working parts. It does not hold the quality discipline that keeps code reliable on a bad connection in a third language. It does not decide what is MVP and what waits for Phase 2. It does not decide what gets cut when the budget pressure arrives, and it will arrive. It cannot do RECON in a real barrio, cannot read whether a community trusts us yet, cannot recruit the first CITI Leader, cannot negotiate the first deal. The build is not only code. It is everything that makes the code worth deploying.
05| What we want: the platform built for a fraction of what a comparable platform cost five years ago, so that the funding ask is small enough to be credible to someone who has never funded infrastructure before.
06| What that requires, by consensus: aggressive AI use on the mechanical layers — generation, scaffolding, tests, docs — with human review, and human leadership on architecture and scope.
07| What we sacrifice: speed bought this way comes with technical debt. Code that appears in an afternoon has not been argued over, and some of it will be rewritten in Phase 2 at a cost that partly cancels the Phase 1 saving. We accept that trade deliberately, because a platform that exists imperfectly beats a perfect one that never gets funded. What we refuse to accept is debt in the places where rework is not possible — the data model, the identity layer, the money paths. Those get built slowly.
08| I am not going to put a productivity multiplier in a book. Two times, five times, ten times — every number I have seen quoted is either a vendor's or a survey of people describing their own performance. The number matters enormously, because the funding ask changes at every multiple, and it is precisely the number I am least qualified to produce. The technical lead sets it, with evidence, before the ask is finalized.
09| The verdict, from where I sit: the platform can be built faster and cheaper than the development sector assumes, and it still cannot be built without a person who decides what gets built, why, and in what order. The build is more affordable than it has ever been and more dependent than ever on the judgment AI cannot supply. Cheaper, not free.
